Stroke, silent cerebral infarct (SCI), transient ischemic attack (TIA) and cognitive impairment cumulatively are frequent and highly morbid complications of sickle cell disease (SCD) in children. Currently, the prevention and treatment of neurological complications of SCD include screening by transcranial Doppler ultrasound (TCD) to identify children at increased risk for strokes, followed by chronic transfusions. Hydroxyurea (HU) reduces the frequency of painful crisis, acute chest syndrome and transfusion and may have beneficial effects on central nervous system (CNS) complications of SCD. A NIH-sponsored phase III trial showed the safety of HU in infants and children. HU for secondary prevention of abnormal TCDs after a period of transfusion is efficacious, but its value in primary prevention of CNS complications of SCD remains unclear, and usage rates early in life remain low. Our preliminary data indicate that a 50% reduction of CNS complications would cause ? 90% of pediatric hematologists to prescribe HU to all young children with SCD. The goal of this project is to plan a definitive primary prevention trial to demonstrate the neuroprotective effect of HU. In a preparatory R34 grant, we con- ducted a feasibility trial showing the acceptability of randomization and the safety of sedation for brain MRIs in young children with SCD. We created the leadership, network of centers and procedures necessary for a defin- itive phase III study (HU Prevent). Our R34 and subsequent RO1 submission used a traditional randomized, double-blind, placebo-controlled study design. Reviews of the RO1 raised concerns about equipoise and whether a placebo would be feasible or informative, as NHLBI guidelines could lead to difficulty convincing physicians and patients to accept randomization. Based on these criticisms, we redesigned our study as a hybrid trial, combining a smaller randomized trial with a non-randomized (observational) cohort. A draft protocol is described in this proposal. The primary outcome is a composite of the weighted risk of impairment due to stroke, SCI, TIA, conditional and abnormal TCD. Secondary outcomes are cognitive function (IQ) and Health Related Quality of Life. Statistical causal inference methods, including propensity score matching, are used to combine patients from the randomized trial and the non-randomized cohort to estimate the overall HU treatment effect using ad- vanced regression models. Subsequent reviews questioned whether the availability of a non-randomized arm could make randomization difficult and create bias, adversely affecting inferences, and the effects of drop-outs or crossovers, given the availability of HU and recent FDA approval of HU for age ? 2 yrs. Given these concerns and the importance of the problem, we feel that a full reassessment of study design, including novel, non-tradi- tional designs is in order, to overcome the issues of sample size in an orphan disease, issues related to random- ization and placebos, difficulty of implementing an optimal control group and potential for crossovers. To inform design, we will use additional electronic phenotype data to refine inclusion and exclusion criteria, assess current HU usage and perform a full community engagement assessment, internal and external to our study consortium.